Genetics has little to say about whether someone starts smoking and quite a lot to say about how heavily they smoke once they do — and, unusually, something practical about which quitting medicine is likely to work better.
Whether a person smokes is mostly a matter of where and when they grew up, who they knew, and what it cost. Genetics barely touches that. What genetics does touch, and firmly, is what happens after someone starts: how many cigarettes a day they settle at, how hard the habit grips, and how difficult it is to stop.
rs1051730, in the nicotinic receptor cluster on chromosome 15 (CHRNA5/CHRNA3), is one of the most solidly replicated findings in behavioural genetics. Each copy of the risk allele is associated with roughly one additional cigarette per day, at p = 3 x 10-73. That is a small number with an enormous certainty behind it — and one cigarette a day, sustained over decades, is not nothing.
The receptor these genes build is the one nicotine binds to. So unlike most associations on this site, the mechanism here is not a mystery: the variant changes the receptor that the drug acts on.
The other two are smaller and cover different ground:
Most gene–drug findings on this site end with a guideline saying not to act on them. This one is different, and it does not come from a variant at all — it comes from a measurement that variants help explain.
Nicotine is cleared by the enzyme CYP2A6, and people clear it at very different rates. That rate can be measured directly in blood as the nicotine metabolite ratio. In a randomised, double-blind trial that assigned people to varenicline, the nicotine patch or placebo after grouping them by that ratio, the result was not the same for both groups: faster metabolisers did better on varenicline than on the patch, while slower metabolisers did about as well on either (Lerman et al., Lancet Respiratory Medicine 2015).
Read that carefully, because it is easy to overstate. It does not say varenicline is better — for slow metabolisers it was not. It says the two are not interchangeable, and which is better depends on something measurable. See the smoking cessation page for what that does and does not mean for a prescription.
It does not say anyone is destined to smoke, or destined to fail at quitting. Cessation rates respond enormously to things that have nothing to do with genotype — cost, bans, support, repeated attempts. Most people who quit do so after several tries, and that pattern holds across every genotype studied.
And nothing here is a reason to delay quitting while working out your genetics. The single most useful fact on this page is that both first-line treatments work; the genetics only refines which one to reach for first.
Phenotypes. "Smoking behaviour" in the GWAS literature covers several distinct traits that are often conflated: initiation (ever versus never), quantity (cigarettes per day), dependence (typically Fagerström-scored), and cessation (former versus current smoker). The loci associated with each differ, and effect estimates are not transferable between them.
CHRNA5/CHRNA3 rs1051730. At 15q25.1, within the nicotinic acetylcholine receptor subunit gene cluster; in strong linkage disequilibrium with the coding variant rs16969968 (CHRNA5 D398N), and the two are frequently reported interchangeably. The GWAS Catalog records a per-allele effect of approximately 1.02 cigarettes per day (p = 3 x 10-73) at a risk allele frequency around 0.65, with separate significant associations for nicotine dependence (p = 6 x 10-20). The locus is additionally associated with lung cancer risk, an association substantially but not entirely mediated through smoking quantity.
BDNF rs6265. The Val66Met coding variant at 11p14.1. Associated with smoking initiation at p = 2 x 10-87 in large meta-analysis with a beta of approximately 0.019 units, and with smoking behaviour at an odds ratio of 1.06 (p = 2 x 10-8). The effect size is small enough that it is of interest for mechanism rather than for individual prediction. The same variant has been examined across a wide range of neuropsychiatric phenotypes with inconsistent replication.
DBH rs3025343. At 9q34.2 near dopamine beta-hydroxylase. Associated specifically with smoking cessation (beta approximately 0.039, p = 4 x 10-12) and with cigarettes per day (p = 6 x 10-11), at an odds ratio of 1.12 for smoking behaviour. Cessation is the less-studied phenotype of the group and effect estimates are correspondingly less settled.
CYP2A6 and treatment selection. Nicotine is metabolised principally by CYP2A6 to cotinine and then to 3'-hydroxycotinine. The nicotine metabolite ratio (3'-hydroxycotinine to cotinine) is a stable phenotypic measure of CYP2A6 activity that captures both genetic and environmental contributions, which is why it is used in preference to genotype alone. Lerman et al. (Lancet Respir Med 2015) randomised smokers to varenicline, nicotine patch or placebo after stratification by NMR: faster metabolisers achieved higher quit rates on varenicline than on the patch, whereas slower metabolisers did not differ significantly between the two active treatments. This is a stratified-trial result rather than a genotype-guided dosing guideline; no regulatory body currently requires NMR or CYP2A6 testing before prescribing.
Clinical status. None of the variants here has a diagnostic or prescribing role. Smoking status and dependence are assessed clinically. The NMR finding is the only element with a plausible route to practice, and it rests on a phenotypic measurement rather than on any of the variants on this page.
What a 23andMe/AncestryDNA export or raw VCF can and can't tell you about Smoking Behaviour and Nicotine Dependence comes down to these specific, well-studied positions — not a diagnosis.
Research-derived gene–drug associations only — not a prescription, dosing guide, or medical advice. Always follow your prescriber's guidance.
| Gene | Drug | What the research shows |
|---|---|---|
| CYP2A6 | Smoking cessation medicines (varenicline, nicotine replacement, bupropion) | The rare pharmacogenomic finding on this site that points at a choice rather than at a dose. CYP2A6 clears nicotine, and how fast a person clears it can be measured in blood as the nicotine metabolite ratio. In a randomised, double-blind trial that grouped smokers by that ratio before assigning treatment, faster metabolisers quit more successfully on varenicline than on the nicotine patch, while slower metabolisers did about as well on either (Lerman et al., Lancet Respiratory Medicine 2015). This is a stratified trial result, not a dosing guideline: no regulator requires the test, and both treatments work. What it argues against is treating the two as interchangeable. (Lerman C et al., Use of the nicotine metabolite ratio as a genetically informed biomarker of response to nicotine patch or varenicline for smoking cessation: a randomised, double-blind placebo-controlled trial. Lancet Respir Med 2015) |
Starting is mostly not. Whether someone takes up smoking tracks era, place, price and peers far more than genotype. What is substantially genetic is what happens next: how many cigarettes a day someone settles at, and how hard the habit is to break.
Because it is per copy of the allele, it is sustained for as long as someone smokes, and it is measured at p = 3 x 10^-73 — one of the most certain findings in the field. Over decades, a persistent difference in daily quantity is not a small exposure difference.
Not from these variants. The useful finding comes from a blood measurement — the nicotine metabolite ratio, which reflects CYP2A6 activity. In a randomised trial, faster metabolisers did better on varenicline than on the patch; slower metabolisers did about as well on either. That is a reason to discuss options with a doctor, not a reason to wait for a test.
No, and this is the part worth being loud about. Quit rates respond enormously to price, bans, support and repeated attempts. Most successful quitters needed several tries, and that is true across every genotype studied. The variants shift a probability; they do not close a door.
Free to reuse. This page's text is original writing from freely-available research, licensed CC BY 4.0 — reuse it, including commercially, with attribution to MyGeneLog. It's general research-derived information, not medical advice or a diagnosis — see Terms of Use.